by greyhaven-ai
a recursive self-improving harness designed to help your agents (and future iterations of those agents) succeed on any task
# Add to your Claude Code skills
git clone https://github.com/greyhaven-ai/autocontextautocontext is a closed-loop control plane for improving agent behavior over repeated runs.
It executes tasks, evaluates outcomes, updates persistent knowledge, and can distill successful behavior into cheaper local runtimes. The goal is to move from frontier-model exploration toward validated, reusable, lower-cost execution.
Most agent systems start every run cold. They do not reliably carry forward what worked, what failed, and what should change next.
autocontext adds that missing feedback loop:
Each generation runs through a structured multi-agent loop:
competitor proposes a strategy or artifact for the taskanalyst explains what happened and whycoach turns that analysis into playbook updates and future hintsarchitect proposes tools, harness improvements, or structural changescurator gates what knowledge is allowed to persistStrategies are then evaluated through scenario execution, staged validation, and gating. Weak changes are rolled back. Successful changes accumulate into reusable knowledge.
No comments yet. Be the first to share your thoughts!
autocontext/ts/.autocontext/docs/agent-integration.md.CONTRIBUTING.md and AGENTS.md.